Long-term agent memory is shifting from predefined workflows toward learnable policies for memory construction and retrieval. Existing learning-based methods…
机构:华为
来源:arXiv 2610.29671 | AI4Papers 论文推荐平台